Marsh headphone amp from Linear Audio

Jump to Latest
#1,361
dabore84 said:
I have a hard time of sourcing the input transistors, what are the equivalent types?
Click to expand...

You can use the somewhat more expensive LSK170 and LSJ74 -- these are from LinearSystems. The pins are oriented differently from the originals.

You can also use the surface mount versions of the 2N5460/2N5457 -- mmbf5460 is $0.49 in single quantities -- just use an SMT to through-hole adapter.
 
#1,362
with LSK170 and LSJ74 you will need some time for offset setting and usually i will take some 10 min (due temco) - after that time amp will warm up and dc offset settle about 1-2mV. From cold amp start - offset will be about 10-15mV. Its better to adjust offset with closed enclosure if possible.
 
#1,365
obscurus said:
LSK170 and LSK74 may need cascode to bring out their best. I used the 2SK170 and 2SJ74 with cascode to bring the Vds to about 6V.
Click to expand...

yes. The originals are low C types so cascade is not needed. Also the originals at the selected operating bias will have close to zero tempco and no drift.... so dc servo also not needed. And finally, the thd will be extreamly low.



THx-Richard
 
#1,367
Cascode will do that on any circuit.... it can also improve CMR. If your supply is not well regulated and filtered and isolated, then cascode will be more needed. Reg and C-Multiplier on PS output for each amp will also take care of any potential problems from low PSRR.



THx-RNMarsh
 
#1,368
jan.didden said:
These guys seem to have it figured out.

Jan
Click to expand...

This isn't 'auto-EQ' that was mentioned by Richard; they need you to listen then calibrate their amp to suit your ears - in other words it's manual not automatic ?

I think Linkwitz did the manual thing himself, he devised some simple series resonant L-C traps for his headphones.
 
#1,369
RNMarsh said:
Cascode will do that on any circuit.... it can also improve CMR. If your supply is not well regulated and filtered and isolated, then cascode will be more needed. Reg and C-Multiplier on PS output for each amp will also take care of any potential problems from low PSRR.



THx-RNMarsh
Click to expand...

Yes, it is. But for this topology, the increasing PSRR is huge and easier and cheaper (at least for me) than make a good Reg and C-Multiplier. It matter of trade-off.
 
#1,370
Bigun said:
This isn't 'auto-EQ' that was mentioned by Richard; they need you to listen then calibrate their amp to suit your ears - in other words it's manual not automatic ?

I think Linkwitz did the manual thing himself, he devised some simple series resonant L-C traps for his headphones.
Click to expand...

Yes, it is not automatic but the difference with the mentioned product is that it does EQ for your personal hearing characteristics. The mentioned product only tried to linearise the headphone response which may actually worsen the listening experience.

Jan
 
#1,371
Personalised EQ - that sounds good. So what does it do exactly, is it a tone generator and a pair of adjustable notch filters ? I remember Linkwitz showing how he listened to his cans with a tone generator to figure out at which frequencies the sound was over emphasized to his hearing and then he notched those frequencies down until the FR sounded right.

I have been interested in building something like this into a preamp project. I have some sensitivity to certain frequencies in the 'presence' region making some recordings far more fatiguing than other people may realise. I had a pair of Grado SR60's that I couldn't live with due to peaks in the FR exciting my hyperacusis which I may have been able to deal with if I had a couple of suitable notch filters.

Perhaps a built-in white noise generator is a simpler option than a swept tone generator. You listen to the noise and adjust a peak filter to find the annoying bits and switch over to a notch. There's gotta be a few ways to do this.
 
Last edited:
#1,372
Richard's design uses a cal run using micro-mikes in your earcanal to get what amounts to the freq response of your ears, and then correct the headphone drive inversely. The idea I believe is to get flat response at the eardrum.

Since an ear is definitely not flat due to all the nooks and crannies, it is doubtfull that flatening the headphone as such will improve the sound. It will surely change it, of course, and that may be enough to make users think it is 'improved'.
I am not an headphone expert but I would think that a headphone tuned for 'best listening experience' would definitely not be flat. Flattening it would be a huge step backwards.

Jan
 
Last edited:
#1,373
Most of proper headphone measurments published in the internet are done with head and torso simulator and are compensated to match ear hearing curve. So flattening headphones accoring to these measurments actually makes them more linear. However there are lots of diffrent target response curves. Curve developed by Harman is probably is most suitable one for hifi headphone listening. This matter is more precisly explained here.
 
#1,374
BamboszeK said:
Most of proper headphone measurments published in the internet are done with head and torso simulator and are compensated to match ear hearing curve. So flattening headphones accoring to these measurments actually makes them more linear. However there are lots of diffrent target response curves. Curve developed by Harman is probably is most suitable one for hifi headphone listening. This matter is more precisly explained here.
Click to expand...

Yes. The measurement is not inside the ear canal near the ear drum. It is outside the ear and measures the headphone response enclosed buy the headphone cup and ear shape/volume. This is similar to a flat response from speakers in a room EQ'ed for flat at the listening location/ear.

This makes the sound 'flat' entering the ear.... then your ear does what it does. It sounds more accurate this way. In this case, the listening is more accurate correction.... it is your ear shape, real flesh, etc and not an average of many or a composite ---- it is Your own ear.



THx-RNMarsh
 
Last edited:
#1,380
Hi Jack,

Ah, great your still around, thanks for replying.

jackinnj said:
I had been mailing them to locations outside the US for $9, but the Post Office squawked because they aren't foldable -- to send outside the US is $13.75.

In respect of Dr. Marsh's IP I sell them at cost which is $3.
Click to expand...

Thanks for still being here and for making boards available at cost, mighty decent of you, especially after five years. Are you still biking the "7 round trip miles to the US Post Office", and is George still there? I'm guessing he's not or you might be able to get away with "non-foldable" boards?

"Foldable", now that's funny, as in Ha Ha, and would make an interesting board design requirement. 🙂 To save you being hassled by George's replacement, I'm really happy to pay the extra postage.

I'd love three boards if that's possible and I'd also really appreciated matched devices for the three boards, if you have any already matched?

Thanks again for still doing this, your a champion! 🙂

I'll send you another PM.

Cheers!
Rob
 
You must log in or register to reply here.
Top Bottom